Merge pull request #130 from Webperf-se/fix/filter-har-to-first-page

Only lint JavaScript belonging to the tested page

Author: marcusosterberg

lib/harAnalyzer.js

@@ -29,6 +29,47 @@ export class HarAnalyzer {
         this.dependencies = this.package.dependencies;
         this.version = this.package.version;
     }
+    getFirstPageEntries(url, harData) {
+        if ('log' in harData) {
+            harData = harData['log'];
+        }
+
+        const entries = harData.entries;
+        if (!Array.isArray(entries)) {
+            return [];
+        }
+
+        // A HAR can contain more than one page, for example when a concurrent
+        // browsertime run ends up in the same browser session (crossed DevTools
+        // port) and navigates to another website mid-recording. Requests made
+        // by other pages must not be attributed to the tested website, and if
+        // the recording doesn't even start with the tested website nothing in
+        // it can be trusted.
+        if (url && entries.length > 0) {
+            const firstUrl = entries[0].request && entries[0].request.url;
+            if (firstUrl) {
+                try {
+                    if (new URL(firstUrl).hostname !== new URL(url).hostname) {
+                        return [];
+                    }
+                } catch {
+                    // Unparsable URLs are handled by the entry loops as before
+                }
+            }
+        }
+
+        const pages = harData.pages;
+        if (!Array.isArray(pages) || pages.length === 0) {
+            return entries;
+        }
+        const firstPageId = pages[0].id;
+        if (firstPageId === undefined) {
+            return entries;
+        }
+        return entries.filter(entry =>
+            entry.pageref === undefined || entry.pageref === firstPageId);
+    }
+
     transform2SimplifiedData(harData, url) {
         const data = {
             'url': url,
@@ -41,13 +82,9 @@ export class HarAnalyzer {
             'script-files': []
         };
 
-        if ('log' in harData) {
-            harData = harData['log'];
-        }
-
         let reqIndex = 1;
 
-        for (const entry of harData.entries) {
+        for (const entry of this.getFirstPageEntries(url, harData)) {
             const req = entry.request;
             const res = entry.response;
             const reqUrl = req.url;